Table A3.
External balancing tests
| African and coloured | ||||||||
|---|---|---|---|---|---|---|---|---|
| Original sample weight | Inverse probability weight | |||||||
|
| ||||||||
| Difference (Teen -Older) | Std. Error |
T-stat | Sign. Diff. |
Difference (Teen -Older) | Std. Error |
T-stat | Sign. Diff. |
|
|
| ||||||||
| Schooling | ||||||||
| Failed grade when 10 | −0.01 | 0.02 | −0.33 | *** | 0.01 | 0.02 | 0.68 | |
| First sex | ||||||||
| Use contraception at first sex | −0.09 | 0.05 | −1.91 | *** | −0.03 | 0.07 | −0.39 | |
| Wave 1 household characteristics | ||||||||
| Wave 1 household per capita income | −38.74 | 83.25 | −0.47 | *** | −22.04 | 92.42 | −0.24 | |
| Room permanent | 0.00 | 0.06 | −0.01 | *** | 0.01 | 0.10 | 0.15 | |
| Flush toilet | 0.00 | 0.04 | −0.08 | *** | −0.01 | 0.06 | −0.12 | |
| Childhood living arrangements | ||||||||
| Lived with mother age 6 | −0.03 | 0.03 | −0.90 | *** | −0.07 | 0.05 | −1.39 | |
| Lived with mother age 7 | −0.02 | 0.03 | −0.46 | *** | −0.06 | 0.05 | −1.10 | |
| Lived with mother age 9 | 0.00 | 0.04 | 0.09 | * | −0.06 | 0.05 | −1.16 | |
| Lived with father age 7 | −0.01 | 0.04 | −0.18 | *** | −0.08 | 0.08 | −1.10 | |
| Lived with father age 8 | 0.00 | 0.05 | 0.09 | * | −0.08 | 0.08 | −1.09 | |
| Lived with father age 9 | −0.01 | 0.04 | −0.33 | *** | −0.11 | 0.08 | −1.39 | |
| Lived with father age 10 | 0.00 | 0.05 | −0.07 | *** | −0.09 | 0.08 | −1.20 | |
| Parental time investment | ||||||||
| Father helps with homework | −0.03 | 0.03 | −1.14 | *** | −0.04 | 0.04 | −1.10 | |
| Mother spent time with just YA at least once a month in past 12 months | −0.06 | 0.05 | −1.16 | *** | −0.04 | 0.08 | −0.53 | |
The table presents the difference in means between teen and older mothers, the standard error of this difference and whether it is significant using the sample weight and the inverse propensity score weight. Variables included are those not included in the propensity score estimation but found to be signficantly different between teen and older mothers when weighted using the sample weight. Other variables not included in the table given that they were not found to be significantly different at base include characteristics of the school – problems, number of learners per class, time taken to school - , failure at other grades, wave 1 asset ownership, living arrangements at other ages since birth and parent time investment – mother helped with homework and discussed personal matters with young adult more than once per month in past 12 months. Differences marked with three asterisks (***) are significant at the 1% level, those marked with two (**) are significant at the 5% level, and those marked with one (*) are significant at the 10% level. The table shows that once the data are weighted by the inverse propensity score weight, differences between teen and older mother characteristics are eliminated or ‘balanced’.